Techniques to reduce the impact of incidents on freeways have been widely implemented by operational freeway agencies. In this article, the techniques of detection, and more specifically on automatic detection through a system of embedded loops, associated detectors, communications and a data processing subsystem are described. Development of the data processing portion of such a system, the incident detection algorithms has been ongoing in the USA since the early 1970's. The seminal work by the Federal Highway Administration is described in IRRD 225375 and 225376. This project produced a set of new algorithms for processing freeway surveillance data to yield indications of the presence of incidents in the freeway traffic stream. Subsequently, a subset of these algorithms were implemented in the Semi-Automated Traffic Management System (SATMS) as a part of the operational freeway surveillance system in Los Angeles. However, the potential of these algorithms was not immediately realized in practice, because the necessary calibrations to determine zone-specific parameters of the algorithms for the zones of the Los Angeles freeways which are under surveillance had not yet been done. This paper describes the ongoing activities of a current research project, the purpose of which has been to determine the appropriate zone-specific parameters. A review is made of the issues entailed in incident detection and a description of the SATMS in Los Angeles and general issues in calibration of incident detection algorithms highlighted. Then the technical approach taken to calibration in the operational environment of SATMS where over 800 zones are involved, and an array of anomalous conditions exist is described. The effort of calibration is ongoing. Interim results are presented here, and representative problems are described. In conclusion an indication of the planned course for the remainder of the calibration effort, is given.
Samenvatting